You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

390 lines
14 KiB

10 years ago
10 years ago
10 years ago
10 years ago
11 years ago
11 years ago
  1. from __future__ import division, unicode_literals
  2. import os
  3. import re
  4. import sys
  5. import time
  6. import random
  7. from ..compat import compat_os_name
  8. from ..utils import (
  9. decodeArgument,
  10. encodeFilename,
  11. error_to_compat_str,
  12. format_bytes,
  13. shell_quote,
  14. timeconvert,
  15. )
  16. class FileDownloader(object):
  17. """File Downloader class.
  18. File downloader objects are the ones responsible of downloading the
  19. actual video file and writing it to disk.
  20. File downloaders accept a lot of parameters. In order not to saturate
  21. the object constructor with arguments, it receives a dictionary of
  22. options instead.
  23. Available options:
  24. verbose: Print additional info to stdout.
  25. quiet: Do not print messages to stdout.
  26. ratelimit: Download speed limit, in bytes/sec.
  27. retries: Number of times to retry for HTTP error 5xx
  28. buffersize: Size of download buffer in bytes.
  29. noresizebuffer: Do not automatically resize the download buffer.
  30. continuedl: Try to continue downloads if possible.
  31. noprogress: Do not print the progress bar.
  32. logtostderr: Log messages to stderr instead of stdout.
  33. consoletitle: Display progress in console window's titlebar.
  34. nopart: Do not use temporary .part files.
  35. updatetime: Use the Last-modified header to set output file timestamps.
  36. test: Download only first bytes to test the downloader.
  37. min_filesize: Skip files smaller than this size
  38. max_filesize: Skip files larger than this size
  39. xattr_set_filesize: Set ytdl.filesize user xattribute with expected size.
  40. (experimental)
  41. external_downloader_args: A list of additional command-line arguments for the
  42. external downloader.
  43. hls_use_mpegts: Use the mpegts container for HLS videos.
  44. http_chunk_size: Size of a chunk for chunk-based HTTP downloading. May be
  45. useful for bypassing bandwidth throttling imposed by
  46. a webserver (experimental)
  47. Subclasses of this one must re-define the real_download method.
  48. """
  49. _TEST_FILE_SIZE = 10241
  50. params = None
  51. def __init__(self, ydl, params):
  52. """Create a FileDownloader object with the given options."""
  53. self.ydl = ydl
  54. self._progress_hooks = []
  55. self.params = params
  56. self.add_progress_hook(self.report_progress)
  57. @staticmethod
  58. def format_seconds(seconds):
  59. (mins, secs) = divmod(seconds, 60)
  60. (hours, mins) = divmod(mins, 60)
  61. if hours > 99:
  62. return '--:--:--'
  63. if hours == 0:
  64. return '%02d:%02d' % (mins, secs)
  65. else:
  66. return '%02d:%02d:%02d' % (hours, mins, secs)
  67. @staticmethod
  68. def calc_percent(byte_counter, data_len):
  69. if data_len is None:
  70. return None
  71. return float(byte_counter) / float(data_len) * 100.0
  72. @staticmethod
  73. def format_percent(percent):
  74. if percent is None:
  75. return '---.-%'
  76. return '%6s' % ('%3.1f%%' % percent)
  77. @staticmethod
  78. def calc_eta(start, now, total, current):
  79. if total is None:
  80. return None
  81. if now is None:
  82. now = time.time()
  83. dif = now - start
  84. if current == 0 or dif < 0.001: # One millisecond
  85. return None
  86. rate = float(current) / dif
  87. return int((float(total) - float(current)) / rate)
  88. @staticmethod
  89. def format_eta(eta):
  90. if eta is None:
  91. return '--:--'
  92. return FileDownloader.format_seconds(eta)
  93. @staticmethod
  94. def calc_speed(start, now, bytes):
  95. dif = now - start
  96. if bytes == 0 or dif < 0.001: # One millisecond
  97. return None
  98. return float(bytes) / dif
  99. @staticmethod
  100. def format_speed(speed):
  101. if speed is None:
  102. return '%10s' % '---b/s'
  103. return '%10s' % ('%s/s' % format_bytes(speed))
  104. @staticmethod
  105. def format_retries(retries):
  106. return 'inf' if retries == float('inf') else '%.0f' % retries
  107. @staticmethod
  108. def best_block_size(elapsed_time, bytes):
  109. new_min = max(bytes / 2.0, 1.0)
  110. new_max = min(max(bytes * 2.0, 1.0), 4194304) # Do not surpass 4 MB
  111. if elapsed_time < 0.001:
  112. return int(new_max)
  113. rate = bytes / elapsed_time
  114. if rate > new_max:
  115. return int(new_max)
  116. if rate < new_min:
  117. return int(new_min)
  118. return int(rate)
  119. @staticmethod
  120. def parse_bytes(bytestr):
  121. """Parse a string indicating a byte quantity into an integer."""
  122. matchobj = re.match(r'(?i)^(\d+(?:\.\d+)?)([kMGTPEZY]?)$', bytestr)
  123. if matchobj is None:
  124. return None
  125. number = float(matchobj.group(1))
  126. multiplier = 1024.0 ** 'bkmgtpezy'.index(matchobj.group(2).lower())
  127. return int(round(number * multiplier))
  128. def to_screen(self, *args, **kargs):
  129. self.ydl.to_screen(*args, **kargs)
  130. def to_stderr(self, message):
  131. self.ydl.to_screen(message)
  132. def to_console_title(self, message):
  133. self.ydl.to_console_title(message)
  134. def trouble(self, *args, **kargs):
  135. self.ydl.trouble(*args, **kargs)
  136. def report_warning(self, *args, **kargs):
  137. self.ydl.report_warning(*args, **kargs)
  138. def report_error(self, *args, **kargs):
  139. self.ydl.report_error(*args, **kargs)
  140. def slow_down(self, start_time, now, byte_counter):
  141. """Sleep if the download speed is over the rate limit."""
  142. rate_limit = self.params.get('ratelimit')
  143. if rate_limit is None or byte_counter == 0:
  144. return
  145. if now is None:
  146. now = time.time()
  147. elapsed = now - start_time
  148. if elapsed <= 0.0:
  149. return
  150. speed = float(byte_counter) / elapsed
  151. if speed > rate_limit:
  152. time.sleep(max((byte_counter // rate_limit) - elapsed, 0))
  153. def temp_name(self, filename):
  154. """Returns a temporary filename for the given filename."""
  155. if self.params.get('nopart', False) or filename == '-' or \
  156. (os.path.exists(encodeFilename(filename)) and not os.path.isfile(encodeFilename(filename))):
  157. return filename
  158. return filename + '.part'
  159. def undo_temp_name(self, filename):
  160. if filename.endswith('.part'):
  161. return filename[:-len('.part')]
  162. return filename
  163. def ytdl_filename(self, filename):
  164. return filename + '.ytdl'
  165. def try_rename(self, old_filename, new_filename):
  166. try:
  167. if old_filename == new_filename:
  168. return
  169. os.rename(encodeFilename(old_filename), encodeFilename(new_filename))
  170. except (IOError, OSError) as err:
  171. self.report_error('unable to rename file: %s' % error_to_compat_str(err))
  172. def try_utime(self, filename, last_modified_hdr):
  173. """Try to set the last-modified time of the given file."""
  174. if last_modified_hdr is None:
  175. return
  176. if not os.path.isfile(encodeFilename(filename)):
  177. return
  178. timestr = last_modified_hdr
  179. if timestr is None:
  180. return
  181. filetime = timeconvert(timestr)
  182. if filetime is None:
  183. return filetime
  184. # Ignore obviously invalid dates
  185. if filetime == 0:
  186. return
  187. try:
  188. os.utime(filename, (time.time(), filetime))
  189. except Exception:
  190. pass
  191. return filetime
  192. def report_destination(self, filename):
  193. """Report destination filename."""
  194. self.to_screen('[download] Destination: ' + filename)
  195. def _report_progress_status(self, msg, is_last_line=False):
  196. fullmsg = '[download] ' + msg
  197. if self.params.get('progress_with_newline', False):
  198. self.to_screen(fullmsg)
  199. else:
  200. if compat_os_name == 'nt':
  201. prev_len = getattr(self, '_report_progress_prev_line_length',
  202. 0)
  203. if prev_len > len(fullmsg):
  204. fullmsg += ' ' * (prev_len - len(fullmsg))
  205. self._report_progress_prev_line_length = len(fullmsg)
  206. clear_line = '\r'
  207. else:
  208. clear_line = ('\r\x1b[K' if sys.stderr.isatty() else '\r')
  209. self.to_screen(clear_line + fullmsg, skip_eol=not is_last_line)
  210. self.to_console_title('youtube-dl ' + msg)
  211. def report_progress(self, s):
  212. if s['status'] == 'finished':
  213. if self.params.get('noprogress', False):
  214. self.to_screen('[download] Download completed')
  215. else:
  216. msg_template = '100%%'
  217. if s.get('total_bytes') is not None:
  218. s['_total_bytes_str'] = format_bytes(s['total_bytes'])
  219. msg_template += ' of %(_total_bytes_str)s'
  220. if s.get('elapsed') is not None:
  221. s['_elapsed_str'] = self.format_seconds(s['elapsed'])
  222. msg_template += ' in %(_elapsed_str)s'
  223. self._report_progress_status(
  224. msg_template % s, is_last_line=True)
  225. if self.params.get('noprogress'):
  226. return
  227. if s['status'] != 'downloading':
  228. return
  229. if s.get('eta') is not None:
  230. s['_eta_str'] = self.format_eta(s['eta'])
  231. else:
  232. s['_eta_str'] = 'Unknown ETA'
  233. if s.get('total_bytes') and s.get('downloaded_bytes') is not None:
  234. s['_percent_str'] = self.format_percent(100 * s['downloaded_bytes'] / s['total_bytes'])
  235. elif s.get('total_bytes_estimate') and s.get('downloaded_bytes') is not None:
  236. s['_percent_str'] = self.format_percent(100 * s['downloaded_bytes'] / s['total_bytes_estimate'])
  237. else:
  238. if s.get('downloaded_bytes') == 0:
  239. s['_percent_str'] = self.format_percent(0)
  240. else:
  241. s['_percent_str'] = 'Unknown %'
  242. if s.get('speed') is not None:
  243. s['_speed_str'] = self.format_speed(s['speed'])
  244. else:
  245. s['_speed_str'] = 'Unknown speed'
  246. if s.get('total_bytes') is not None:
  247. s['_total_bytes_str'] = format_bytes(s['total_bytes'])
  248. msg_template = '%(_percent_str)s of %(_total_bytes_str)s at %(_speed_str)s ETA %(_eta_str)s'
  249. elif s.get('total_bytes_estimate') is not None:
  250. s['_total_bytes_estimate_str'] = format_bytes(s['total_bytes_estimate'])
  251. msg_template = '%(_percent_str)s of ~%(_total_bytes_estimate_str)s at %(_speed_str)s ETA %(_eta_str)s'
  252. else:
  253. if s.get('downloaded_bytes') is not None:
  254. s['_downloaded_bytes_str'] = format_bytes(s['downloaded_bytes'])
  255. if s.get('elapsed'):
  256. s['_elapsed_str'] = self.format_seconds(s['elapsed'])
  257. msg_template = '%(_downloaded_bytes_str)s at %(_speed_str)s (%(_elapsed_str)s)'
  258. else:
  259. msg_template = '%(_downloaded_bytes_str)s at %(_speed_str)s'
  260. else:
  261. msg_template = '%(_percent_str)s % at %(_speed_str)s ETA %(_eta_str)s'
  262. self._report_progress_status(msg_template % s)
  263. def report_resuming_byte(self, resume_len):
  264. """Report attempt to resume at given byte."""
  265. self.to_screen('[download] Resuming download at byte %s' % resume_len)
  266. def report_retry(self, err, count, retries):
  267. """Report retry in case of HTTP error 5xx"""
  268. self.to_screen(
  269. '[download] Got server HTTP error: %s. Retrying (attempt %d of %s)...'
  270. % (error_to_compat_str(err), count, self.format_retries(retries)))
  271. def report_file_already_downloaded(self, file_name):
  272. """Report file has already been fully downloaded."""
  273. try:
  274. self.to_screen('[download] %s has already been downloaded' % file_name)
  275. except UnicodeEncodeError:
  276. self.to_screen('[download] The file has already been downloaded')
  277. def report_unable_to_resume(self):
  278. """Report it was impossible to resume download."""
  279. self.to_screen('[download] Unable to resume')
  280. def download(self, filename, info_dict):
  281. """Download to a filename using the info from info_dict
  282. Return True on success and False otherwise
  283. """
  284. nooverwrites_and_exists = (
  285. self.params.get('nooverwrites', False) and
  286. os.path.exists(encodeFilename(filename))
  287. )
  288. if not hasattr(filename, 'write'):
  289. continuedl_and_exists = (
  290. self.params.get('continuedl', True) and
  291. os.path.isfile(encodeFilename(filename)) and
  292. not self.params.get('nopart', False)
  293. )
  294. # Check file already present
  295. if filename != '-' and (nooverwrites_and_exists or continuedl_and_exists):
  296. self.report_file_already_downloaded(filename)
  297. self._hook_progress({
  298. 'filename': filename,
  299. 'status': 'finished',
  300. 'total_bytes': os.path.getsize(encodeFilename(filename)),
  301. })
  302. return True
  303. min_sleep_interval = self.params.get('sleep_interval')
  304. if min_sleep_interval:
  305. max_sleep_interval = self.params.get('max_sleep_interval', min_sleep_interval)
  306. sleep_interval = random.uniform(min_sleep_interval, max_sleep_interval)
  307. self.to_screen(
  308. '[download] Sleeping %s seconds...' % (
  309. int(sleep_interval) if sleep_interval.is_integer()
  310. else '%.2f' % sleep_interval))
  311. time.sleep(sleep_interval)
  312. return self.real_download(filename, info_dict)
  313. def real_download(self, filename, info_dict):
  314. """Real download process. Redefine in subclasses."""
  315. raise NotImplementedError('This method must be implemented by subclasses')
  316. def _hook_progress(self, status):
  317. for ph in self._progress_hooks:
  318. ph(status)
  319. def add_progress_hook(self, ph):
  320. # See YoutubeDl.py (search for progress_hooks) for a description of
  321. # this interface
  322. self._progress_hooks.append(ph)
  323. def _debug_cmd(self, args, exe=None):
  324. if not self.params.get('verbose', False):
  325. return
  326. str_args = [decodeArgument(a) for a in args]
  327. if exe is None:
  328. exe = os.path.basename(str_args[0])
  329. self.to_screen('[debug] %s command line: %s' % (
  330. exe, shell_quote(str_args)))